Re-architecting datacenter networks and stacks for low latency and high performance

Handley, Mark; Raiciu, Costin; Agache, Alexandru; Voinescu, Andrei; Moore, Andrew W.; Antichi, Gianni; Wójcik, Marcin

doi:10.1145/3098822.3098825

Cited by 280 publications

(191 citation statements)

References 34 publications

(41 reference statements)

Supporting

Mentioning

176

Contrasting

Order By: Relevance

“…Naturally, there is a continuous flow of systems papers proposing new networking architectures, e.g. for SDN [17], data center networks [48], content delivery networks [24] or cloud computing [106], to name a few. Yet, we are unaware of any system-level papers proposing a quantum network stack including protocols for concrete hardware implementations.…”

Section: Related Workmentioning

confidence: 99%

A link layer protocol for quantum networks

Dahlberg

Skrzypczyk

Coopmans

et al. 2019

Proceedings of the ACM Special Interest Group on Data Communication

193

269

View full text Add to dashboard Cite

Quantum communication brings radically new capabilities that are provably impossible to attain in any classical network. Here, we take the first step from a physics experiment to a fully fledged quantum internet system. We propose a functional allocation of a quantum network stack and construct the first physical and link layer protocols that turn ad-hoc physics experiments producing heralded entanglement between quantum processors into a well-defined and robust service. This lays the groundwork for designing and implementing scalable control and application protocols in platform-independent software. To design our protocol, we identify use cases, as well as fundamental and technological design considerations of quantum network hardware, illustrated by considering the state-of-the-art quantum processor platform available to us (Nitrogen-Vacancy (NV) centers in diamond). Using a purpose built discrete-event simulator for quantum networks, we examine the robustness and performance of our protocol using extensive simulations on a supercomputing cluster. We perform a full implementation of our protocol, where we successfully validate the physical simulation model against data gathered from the NV hardware. We first observe that our protocol is robust even in a regime of exaggerated losses of classical control messages with only little impact on the performance of the system.We proceed to study the performance of our protocols for 169 distinct simulation scenarios, including tradeoffs between traditional performance metrics such as throughput and the quality of entanglement. Finally, we initiate the study of quantum network scheduling strategies to optimize protocol performance for different use cases. 1 arXiv:1903.09778v1 [quant-ph]

show abstract

Section: Related Workmentioning

confidence: 99%

A link layer protocol for quantum networks

Dahlberg

Skrzypczyk

Coopmans

et al. 2019

Proceedings of the ACM Special Interest Group on Data Communication

193

269

View full text Add to dashboard Cite

show abstract

“…The number of priority levels would depend on the network and application. For instance, in the data center, it is sometimes beneficial to cut the payload and only forward the header [7,16]. This is due to the use of shallow buffers in order to speed up communications.…”

Section: Packet Formatmentioning

confidence: 99%

“…Moreover, Packet Wash is a novel "significance-based" scheme and works at the packet level, not at bit-level encoding. With respect to the trimming method, specific to data centers, [7] and [16], are receiver-driven traffic control mechanisms that use packet trimming. In particular, the goal is to achieve fast retransmissions; therefore, nodes have very shallow buffers.…”

Section: Related Workmentioning

confidence: 99%

“…This results in unpredictable delays as well as an increase in the network load, wasting network resources/capacity. To mitigate this problem, different schemes have been proposed such as in data centers [7,16,25], media streaming [12,13], and wireless networks [3]. While some of these schemes are based on mechanisms for efficient and faster re-transmissions, and others utilize redundant transmissions, we propose a novel approach that attempts to eliminate or at least effectively reduce the re-transmissions in the network.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Framework for Qualitative Communications Using Big Packet Protocol

Makhijani

Yousefi

et al. 2019

Proceedings of the ACM SIGCOMM 2019 Workshop on Networking for Emerging Applications and Technologies

View full text Add to dashboard Cite

In the current Internet architecture, a packet is a minimal or fundamental unit upon which different actions such as classification, forwarding, or discarding are performed by the network nodes. When faced with constrained or poor network conditions, a packet is subjected to undesirable drops and re-transmissions, resulting in unpredictable delays and subsequent traffic overheads in the network. Alternately, we introduce qualitative communication services which allow partial, yet timely, delivery of a packet instead of dropping it entirely. These services allow breaking down packet payloads into smaller units (called chunks), enabling much finer granularity of bandwidth utilization.We propose Packet Wash as a new operation in forwarding nodes to support qualitative services. Upon packet error or network congestion, the forwarding node selectively removes some chunk(s) from the payload based on the relationship among the chunks or the individual significance level of each chunk. We also present a qualitative communication framework as well as a Packet Wash directive implemented in a newly evolved data plane technology, called Big Packet Protocol (BPP).

show abstract

“…(3) NUMFabric [40] achieves more flexible and faster bandwidth allocation than TCP but still employs iterative convergence (e.g., 31 RTTs). And, (4) while ExpressPass [8] and NDP [23] target general congestion via receiver-based congestion control, neither scheme isolates receiver congestion. ExpressPass employs BIC-TCP iterative convergence which takes 20 RTTs for a datacenter network (Section 5.1); ExpressPass shows results only for a simple network.…”

Section: Introductionmentioning

confidence: 99%

Dart: Divide and Specialize for Fast Response to Congestion in RDMA-Based Datacenter Networks

Xue

Chaudhry

Vamanan

et al. 2020

IEEE/ACM Trans. Networking

View full text Add to dashboard Cite

Though Remote Direct Memory Access (RDMA) promises to reduce datacenter network latencies significantly compared to TCP (e.g., 10x), end-to-end congestion control in the presence of incasts is a challenge. Targeting the full generality of the congestion problem, previous schemes rely on slow, iterative convergence to the appropriate sending rates (e.g., TIMELY takes 50 RTTs). Several papers have shown that even in oversubscribed datacenter networks most congestion occurs at the receiver. Accordingly, we propose a divide-and-specialize approach, called Dart, which isolates the common case of receiver congestion and further sub-divides the remaining in-network congestion into the simpler spatially-localized and the harder spatially-dispersed cases. For receiver congestion, we propose direct apportioning of sending rates (DASR) in which a receiver for n senders directs each sender to cut its rate by a factor of n, converging in only one RTT. For the spatially-localized case, Dart provides fast (under one RTT) response by adding novel switch hardware for in-order flow deflection (IOFD) because RDMA disallows packet reordering on which previous load balancing schemes rely. For the uncommon spatially-dispersed case, Dart falls back to DCQCN. Small-scale testbed measurements and at-scale simulations, respectively, show that Dart achieves 60% (2.5x) and 79% (4.8x) lower 99 th -percentile latency, and similar and 58% higher throughput than InfiniBand, and TIMELY and DCQCN.

show abstract

Re-architecting datacenter networks and stacks for low latency and high performance

Cited by 280 publications

References 34 publications

A link layer protocol for quantum networks

A link layer protocol for quantum networks

A Framework for Qualitative Communications Using Big Packet Protocol

Dart: Divide and Specialize for Fast Response to Congestion in RDMA-Based Datacenter Networks

Contact Info

Product

Resources

About